Calculating the cost of programming is the most critical component used to measure effort and time to accomplish the task during software project management. Nowadays, there are several models for cost forecasting program development, supplying useful information to project management team to make the best choices. One such notable numerical model is the COCOMO model. The objective is to modify the coefficients of intermediate COCOMO model with a meta-heuristic algorithm. Estimated cost will be compared with the real data. The methodology is applied on COCOMO NASA software dataset.